Messina, S., Eens, M., Casasole, G., AbdElgawad, H., Asard, H.,

Pinxten, R. and Costantini, D. (2017) Experimental inhibition of a key

cellular antioxidant affects vocal communication. Functional Ecology,

31, pp. 1101-1110. (doi:10.1111/1365-2435.12825)

There may be differences between this version and the published

version. You are advised to consult the publisher’s version if you wish

to cite from it.

This is the peer-reviewed version of the following article: Messina, S.,

Eens, M., Casasole, G., AbdElgawad, H., Asard, H., Pinxten, R. and

Costantini, D. (2017) Experimental inhibition of a key cellular

antioxidant affects vocal communication. Functional Ecology, 31, pp.

1101-1110, which has been published in final form at 10.1111/1365-

2435.12825. This article may be used for non-commercial purposes in

accordance with Wiley Terms and Conditions for Self-Archiving.

http://eprints.gla.ac.uk/139116/

Deposited on: 31 March 2017

Enlighten – Research publications by members of the University of Glasgow

http://eprints.gla.ac.uk

Experimental inhibition of a key cellular antioxidant affects vocal communication

Simone Messina1, Marcel Eens

1, Giulia Casasole

1, Hamada AbdElgawad

2,3, Han

Asard2, Rianne Pinxten

1,4, David Costantini

1,5,6*

1Behavioural Ecology & Ecophysiology group, Department of Biology, University of

Antwerp, Universiteitsplein 1, 2610 Wilrijk, Belgium; 2Integrated Molecular Plant

Physiology Research, Department of Biology, University of Antwerp, Antwerp,

Belgium; 3Department of Botany, Faculty of Science, University of Beni-Suef, Beni-Suef

62511, Egypt; 4Faculty of Social Sciences, Antwerp School of Education, University of

Antwerp, Antwerp, Belgium; 5Institute of Biodiversity, Animal Health and Comparative

Medicine, University of Glasgow, Glasgow G12 8QQ, UK; 6Current address:

Department of Evolutionary Ecology, Leibniz Institute for Zoo and Wildlife Research,

Alfred-Kowalke-Str. 17, 10315 Berlin, Germany.

* Corresponding author: davidcostantini@libero.it

Running headline: Song rate and oxidative stress

2

Summary

1. There is substantial interest of evolutionary ecologists in the proximate 1

mechanisms that modulate vocal communication. In recent times, there has been 2

growing interest in the role of oxidative stress as a mediator of avian song 3

expression. 4

2. Here we tested whether the experimental inhibition of the synthesis of a key 5

cellular antioxidant (glutathione) reduces song rate metrics of male European 6

starlings (Sturnus vulgaris). We measured the effect of our treatment on total song 7

rate and on its two components, undirected and nest-box oriented song, outside 8

the breeding season. 9

3. Treated males that did not own a nest-box (subordinate males likely to be of lower 10

quality) suffered increased oxidative stress relative to untreated males, while 11

treated males that owned a nest-box (dominant males likely to be of higher 12

quality) did not. Treated non-owners also reduced their undirected song rate, 13

whereas treated nest-box owners did not suffer any reduction in song rate. 14

4. Our results revealed that inhibition of a key cellular antioxidant results in 15

decreased vocal communication in a social vertebrate, and that this effect is 16

dependent on its social status (nest-box owner versus non-owner). 17

5. This work provides support for the hypothesis that acoustic signals may honestly 18

convey information about the individual oxidative status and capacity to regulate 19

the oxidative balance. Our findings raise the possibility of hitherto unexplored 20

impacts of oxidative stress on fitness traits in social species. 21

22

3

Key-words: antioxidants, constraint, glutathione, honest-signalling, life history, 23

oxidative damage, sexual selection, song rate, Sturnus vulgaris 24

4

Introduction 25

Production of visual and non-visual signals is a key component of social interaction. 26

Sounds, conspicuous colourations and odours are some of the most notable ways 27

animals use to communicate with conspecifics. The expression of these signals largely 28

varies among individuals. One reason for this variation lies with the individual quality, 29

where high quality individuals are expected to express more exaggerated signals than 30

low quality individuals. This is because either production or maintenance of honest 31

signals carry costs that only high quality individuals would be able to afford (Lindström 32

et al. 2009; Pike et al. 2010; Garratt & Brooks 2012). 33

Acoustic signals are condition-dependent in a variety of invertebrates and 34

vertebrates and advertise to conspecifics their social or mating status (e.g., Hunt et al. 35

2004; Ball et al. 2006; Koren 2006; Humfeld 2013). Avian song is a renowned acoustic 36

trait that may convey several attributes of individual quality (Gil & Gahr 2002). For 37

example, it has been found that different traits of song output are related to immune 38

function (Duffy & Ball 2002), to food availability and body mass (Ritschard & Brumm 39

2012) and to stress response and survival (MacDougall-Shackleton et al. 2009). Avian 40

song is also linked with dominance and territory ownership. For example, male 41

European starlings Sturnus vulgaris that own a nest-box sing at higher rates compared 42

to individuals that do not own a nest-box (Kelm-Nelson et al. 2011; DeVries et al. 43

2016). 44

There are clearly constraints that limit song expression. For example, it has been 45

shown that immunization can reduce song rate in collared flycatchers Ficedula 46

albicollis (Garamszegi et al. 2004) and in white-browed sparrow weavers (York et al. 47

2016), or rattle duration in barn swallows Hirundo rustica (Dreiss et al. 2008). Recent 48

5

studies in male European starlings found that inflammatory processes significantly 49

decreased song rate (Casagrande et al. 2015) and that the antibody production caused a 50

moderate reduction of one particular mode of singing (undirected song rate, i.e. song 51

produced away from the nest-box; Costantini et al. 2015). Song rate can also be 52

constrained by sex steroid hormones. Experimental manipulations of testosterone in 53

male European starlings showed that this hormone can influence song rate (De Ridder, 54

Pinxten & Eens 2000; Pinxten et al. 2002; Ball et al. 2006; Van Hout et al. 2012). 55

Finally, there is also evidence that the size of certain song nuclei limit the expression of 56

song (reviewed in Garamszegi & Eens 2004). 57

Which cellular mechanisms may constrain signalling has been a central question 58

in the study of animal communication in recent years (Hill 2011). In this regard, one 59

cellular mechanism thought to be particularly important is oxidative stress (von Schantz 60

et al. 1999; Garratt & Brooks 2012; Casagrande, Pinxten & Eens 2016), a complex 61

biochemical condition of the organism that is dependent on the rate of oxidative damage 62

generation and oxidation of non-protein and protein thiols that regulate the cell 63

oxidative balance (Jones 2006; Halliwell & Gutteridge 2007; Costantini & Verhulst 64

2009; Sohal & Orr 2012). Dysfunctional regulation of the oxidative balance might be a 65

significant handicap for the expression of sexual signals in low quality individuals 66

(Garratt & Brooks 2012). 67

Recent studies found significant correlations between song rate and some 68

metrics of either oxidative damage or non-enzymatic antioxidants in European starlings 69

and snow buntings (Plectrophenax nivalis) (reviewed in Casagrande, Pinxten & Eens 70

2016). Although these studies suggest that there might be a link between song 71

production and oxidative stress, the role of oxidative stress as a constraint on song 72

6

production and, hereby, the role of song as a signal of oxidative status, has never been 73

experimentally tested. 74

In this study, we tested whether a change in individual oxidative balance through 75

experimental inhibition of the synthesis of a key cellular antioxidant (glutathione; Jones 76

2006) reduces song rate metrics of male European starlings. We also tested whether the 77

role of oxidative stress as a constraint on song rate differs between birds that either own 78

or do not own a nest-box. Male European starlings are an excellent study system to 79

address these questions. In male starlings the acquisition of a nest site/nest-box is a 80

good indicator of individual quality and dominance status. Males that acquire a nest-box 81

chase other males from perches or feeding sites more frequently and exhibit higher song 82

rate on or inside the nest-box (i.e., nest-box oriented song rate) than other males (Eens 83

1997; Riters et al. 2000; Spencer et al. 2004; Sartor & Ball 2005; Kelm-Nelson et al. 84

2011; Cordes et al. 2014; DeVries et al. 2016). The nest-box oriented song rate is also 85

used by starling males outside the breeding season for the acquisition and defense of a 86

nest site (Gwinner, Gwinner & Dittami 1987; Eens 1997). European starlings also sing 87

away from the nest-box throughout the year; this undirected song is used for 88

maintaining group cohesion and social order (Eens 1997). It has been shown that song 89

rate metrics in this species are associated with levels of antioxidants and oxidative 90

damage during both the breeding and non-breeding season (Van Hout, Eens & Pinxten 91

2011; Casagrande et al. 2014; Costantini et al. 2015). In contrast to most other songbird 92

species, starling males sing at high levels throughout most of the year (apart from the 93

moulting period; Eens 1997; Riters et al. 2000; Van Hout et al. 2009). 94

95

Materials and Methods 96

7

HOUSING CONDITIONS 97

Adult male starlings used for the experiment had been previously captured in the 98

Antwerp district and housed in large single-sex outdoor aviaries on the grounds of 99

Campus Drie Eiken of the University of Antwerp (Wilrijk, Belgium). At the start of the 100

experiment, all starlings had been kept in captivity for at least one year and were 101

between 2 and 9 years old. 102

To exclude any potential confounding effect of testosterone, we conducted our 103

study during the fall season when testosterone levels are basal, as indicated by the black 104

beak colouration of our experimental birds, and when all birds had completed the moult 105

(Eens 1997; Riters et al. 2000). On October 26, 2015, 32 starlings were moved in two 106

adjacent outdoor aviaries (L × W × H; 27.0 × 7.0 × 2.75 m). Birds were randomly 107

allocated to the two aviaries (16 birds in each aviary). In each aviary, there were both 108

control birds and birds going to be treated (8 control and 8 treated birds in aviary A and 109

8 control and 8 treated birds in aviary B). However, one control bird from aviary A died 110

before the start of the treatment, limiting the number of birds used in that aviary to 15 111

(and 31 in total). Each aviary had 16 nest-boxes and each nest-box had a perch in front 112

of it. All starlings were marked with a unique combination of coloured bands and a 113

metal ring, which allowed individual recognition. Food (Orlux, Deinze, Belgium; Nifra 114

Van Camp, Boechout, Belgium) and water were provided ad libitum. 115

116

EXPERIMENTAL DESIGN 117

The experimental treatment started about four weeks after the starlings were moved in 118

the new aviaries in order to allow them to get accustomed with the new environment. 119

The experiment was performed according to the timescale in Fig. 1. During 12-21 120

8

November, the song of all males was scored and males were also identified as owners or 121

non-owners of a nest-box (see below). On November 23, the manipulation of individual 122

oxidative balance was started. Starlings were given intramuscular injections of 200 μl of 123

a solution containing 50 mg of DL-buthionine-(S,R)-sulfoximine (Sigma-Aldrich 124

B2640) per ml of saline solution (PBS). The birds were injected five times on alternate 125

days (Fig. 1). The control individuals were subjected to the same regime, but were 126

injected with PBS only. Further details are given in the paragraph “Glutathione 127

synthesis inhibition”. 128

Immediately before the first injection and the day after the last injection, at a 129

comparable time of day (13:30-15:00), a sample of blood (ca. 500 μl) was collected by 130

venipuncture of the wing vein using heparinized microvettes (Sarstedt, Nümbrecht, 131

Germany). Body mass was also recorded. Blood samples were maintained at around 132

+4°C while on the field. When back in the laboratory, tubes were centrifuged in order to 133

separate plasma from red blood cells. Both plasma and red blood cell sample were 134

stored at -80 °C. In order to capture starlings, we took advantage of the small (entrance) 135

aviary attached to each of the two large outdoor aviaries. On the capture day, the door of 136

this small aviary was opened and then two of us, standing on the opposite side of the 137

aviary, made the birds fly into the small aviary. Inside this small aviary, it was easy to 138

capture birds with a butterfly net without causing any damage to them. Some birds were 139

also captured using a butterfly net while they were flying to the small aviary. The whole 140

procedure took a maximum of 10 minutes for each of the two aviaries. 141

142

QUANTIFICATION OF SONG RATE AND OWNERSHIP 143

9

All behavioural observations were made by the same person (using a binocular) from 144

behind a shelter located ca. 5 m from each aviary. Using a one-zero sampling technique 145

(Martin & Bateson 2007), we monitored the behaviour of all starlings within one aviary, 146

in uninterrupted sessions of 60 minutes, between 10h00 and 12h00 (when singing 147

activity of starlings during the day is highest, Eens 1997). Every minute, the aviary was 148

scanned from the left to the right side to register which males were singing. All the 149

observations were registered using an automatic voice recorder, which enabled us to 150

keep looking at the birds without any distractions. We alternated the order of the 151

aviaries between subsequent days in order to have a balanced distribution of the timing 152

of observations. Behavioural observations were made for 10 consecutive days (12-21 153

November) before birds were given the first injection. The average value of these data 154

was used as pre-treatment value. Thereafter, the collection of behavioural data 155

continued from the day after the first injection until the third day after the last injection 156

(Fig. 1). On days of injections, the song behaviour of the males was always registered 157

before the injections. During all behavioural observations, both the nest-box oriented 158

song rate and the undirected song rate were scored for each individual. Given that 159

European starlings, while singing, adopt a characteristic upright stance, with an 160

upturned bill and the throat feathers and beak can be seen moving (Fear 1984), singing 161

behaviour can be easily quantified. It is also easy to score males singing inside a nest-162

box because they sing with their head looking out from the nest-box hole, which makes 163

them visible. The ring of birds singing inside a nest-box was either identified while the 164

bird entered the nest-box or while it left it. The nest-box oriented song rate was 165

quantified as the proportion of samples per session during which a male was singing 166

while inside a nest-box, on the top of it or on the perch connected to the nest-box. The 167

10

undirected song rate was quantified as the proportion of samples per session when a 168

male was singing away from nest-boxes, for example on perches away from nest-boxes 169

or on the ground (Pinxten et al. 2002; Casagrande et al. 2015; Costantini et al. 2015). 170

Based on behavioural observations conducted before the treatment (12-21 November), 171

birds were also classified as either owners or non-owners of a nest-box. Owners were 172

birds observed occupying the same nest-box (singing at or in it and repeatedly entering 173

and leaving it) and chasing other males away from it (Gwinner, Van’t Hof & Zeman 174

2002; Spencer et al. 2004; Kelm-Nelson, Stevenson & Riters 2012) for at least 8 days 175

out of the 10 observation days. In case birds were seen to sing on the roof of a nest-box 176

but did not perform other ownership-linked behaviours, they were considered as non-177

owners. All the birds that were considered owners before the start of the treatment were 178

still owners in the period after the start of the treatment (observed singing, entering and 179

leaving the same nest-box and chasing other males away from it for at least 10 out of 12 180

days). Note that we relied on a rather conservative definition of nest-box ownership, 181

implying that non-owners may also have occupied a nest-box (and hence produced nest-182

box oriented song) during a short period. Male captive European starlings vigorously 183

defend a nest-box not only during the breeding season, but also after the moult has been 184

completed from September onwards. Although abundant nest-boxes were provided in 185

the aviaries, not all males became nest-box owners because some males occupied and 186

defended more than one nest-box. The captive set-up enabled us to distinguish between 187

dominant and subdominant males. Social status (in the form of nest-box possession) has 188

previously been shown to affect song rate (e.g., Eens 1997; Riters et al. 2000, 2012, 189

2014). 190

191

11

GLUTATHIONE SYNTHESIS INHIBITION 192

To induce a state of oxidative stress, starlings were given intramuscular injections of 193

200 μl of a pure solution containing 50 mg of DL-buthionine-(S,R)-sulfoximine (Sigma-194

Aldrich B2640) per ml of saline solution (PBS). Sulfoximine is a non-toxic drug that 195

reduces the synthesis of glutathione, a key cellular antioxidant, by inhibiting the activity 196

of the enzyme gamma-glutamylcysteine synthetase (Griffith & Meister 1979; Griffith 197

1982; Galván & Alonso-Alvarez 2008; Costantini et al. 2016; Koch & Hill 2016). The 198

injections were done on alternate days for a total of five times (Fig. 1). The total amount 199

of sulfoximine given to each treated individual corresponded to 50 mg; this amount was 200

chosen based on work with other songbird species (Galván & Alonso-Alvarez 2008; 201

Romero-Haro & Alonso-Alvarez 2015; Costantini et al. 2016). This resulted in variable 202

doses (i.e., amount of sulfoximine per gram of body mass) given that there was 203

variation in body mass among birds (range of body mass of treated individuals from 75 204

to 95 grams), but the body mass did not differ significantly between treatment groups 205

nor was it affected by the treatment (see Table 1). The control individuals were 206

subjected to the same regime, but injected with PBS only. 207

208

ANALYSIS OF BLOOD OXIDATIVE STATUS 209

To validate the effect of treatment on glutathione concentration and oxidative damage, 210

two methods commonly applied to vertebrates were used. First, high-performance liquid 211

chromatography with electrochemical detection was applied for simultaneous 212

determination of reduced glutathione (GSH, tripeptide synthetized by the organism that 213

has antioxidant properties) and oxidized glutathione (GSSG, it is the molecule of GSH 214

that has been oxidised, for example, after reaction with a free radical) in red blood cells 215

12

by a Reversed-Phase HPLC of Shimadzu (Hai Zhonglu, Shanghai). We applied the 216

protocol as described in Sinha et al. (2014). Concentrations of GSH and GSSG were 217

expressed as µmol g-1

fresh weight of red blood cells. We calculated the GSH/GSSG 218

ratio that was used as an index of redox state (higher values indicate lower oxidative 219

stress; Jones 2006). Second, the d-ROMs assay (Reactive Oxygen Metabolites; Diacron 220

International, Grosseto, Italy) was used to measure plasma oxidative damage 221

metabolites (mostly organic hydroperoxides) that are generated early in the oxidative 222

cascade. Our work relied on this metric of oxidative damage because GSH is used to 223

detoxify the organism from organic hydroperoxides (Halliwell & Gutteridge 2007). 224

Hence, a decrease of GSH is expected to result in an increase of ROMs. Analyses of 225

ROMs were done according to manufacturer’s instructions as in previous studies (e.g. 226

Costantini et al. 2015). Quality controls (Diacron International) were also assessed in 227

each assay. Values of ROMs have been expressed as mM of H2O2 equivalents. 228

229

STATISTICAL ANALYSES 230

Linear mixed models with a repeated measures design (SAS Version 9.3, Cary, NC, 231

USA) were used to assess the effect of treatment on reduced glutathione (GSH), 232

oxidised glutathione (GSSG), the GSH/GSSG ratio, plasma reactive oxygen metabolites 233

and body mass. In each full model, treatment group (control and treated), ownership 234

(owner and non-owner of a nest-box), sampling day (pre- and post-treatment) and all 235

their interactions were included as fixed factors; aviary and individual nested within 236

aviary were both entered as random factors. Pre- and post-treatment values of oxidative 237

status metrics refer to blood samples collected on 23 November and 2 December, 238

respectively (Fig. 1). Non-significant interactions were sequentially removed starting 239

13

from the three-way interaction; main factors were always retained in the reduced models 240

irrespective of their statistical significance. Post-hoc Tukey tests were used to explore 241

further any significant interactions. 242

Generalized linear mixed models with a binomial error distribution (lmer in 243

package ‘lme4’, R version 3.1.1; R Core Team, 2013) were used to test the effect of 244

treatment on total song rate, nest-box oriented song rate and undirected song rate, 245

respectively. We relied on binomial models because we had count data (i.e., number of 246

times a bird was seen to sing over 60 minutes) that were bounded (i.e., ranging from 0 247

to 60), hence the use of percentages as response variable raises a few concerns (chapter 248

16 in Crawley 2007). For graphical purposes, we have, however, shown data of each 249

song rate metric as percentages. In each full model, treatment groups (control and 250

treated), ownership (owner and non-owner of a nest-box), sampling day (13 days in 251

total) and all their interactions were included as fixed factors; aviary and individual 252

nested within aviary were both entered as random factors. To account for 253

overdispersion, which is common in Poisson models, an observation level random effect 254

was also added (Harrison 2014). As pre-treatment (i.e., sampling day 1) values of each 255

song metric, we used the average of the values recorded from 12 to 21 November (Fig. 256

1), i.e., before the first injection. The following twelve days of collection of song rate 257

were those from 23 November to 5 December (Fig. 1). Non-significant interactions 258

were sequentially removed starting from the three-way interaction; main factors were 259

always retained in the reduced models irrespective of their statistical significance. Post-260

hoc tests were used to explore further any significant interactions. 261

Linear mixed models were also used to assess whether any of the song rate 262

metrics was associated with d-ROMs values. To do so, we pooled all males together 263

14

because the sample size was not adequate to test covariation within each treatment 264

group. Within each model, both pre- and post-treatment values of d-ROMs were pooled 265

together and were included as a single fixed predictor; aviary and individual nested 266

within aviary were both entered as random factors because pre- and post-treatment 267

values are not independent from each other. Of each song metric, we included within 268

each model the pre-treatment values (those recorded on day 1, see Figure 1) together 269

with values recorded on sampling day 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12 or 13, respectively. 270

Preliminary linear models showed that at the time of the first injection control 271

and treated owner or non-owner birds were similar for age, body mass and tarsus length 272

in both aviaries (treatment group × ownership × aviary, p-values ≥ 0.24). In the two 273

aviaries, the number of owners (8 each) and non-owners (7 and 8, respectively) was 274

similar, as well as the proportion of owners and non-owners was evenly distributed in 275

control and treated birds. All the birds that were considered owners before the start of 276

the treatment were still owners in the period after the start of the treatment. 277

278

Results 279

Irrespective of ownership, the concentration of GSH in red blood cells increased 280

significantly in control individuals, while it was stable in treated birds (Table 1, Fig. 2). 281

This indicates that sulfoximine prevented upregulation of GSH synthesis. The 282

concentration of GSSG in red blood cells and the GSH/GSSG ratio respectively 283

increased (estimate±SE: 0.48±0.10) and decreased (estimate±SE: -1.44±0.46) with time 284

in all birds, irrespective of treatment group and ownership (Table 1, Fig. 2). The 285

difference in plasma ROMs between control and treated individuals depended on 286

whether or not a bird was owner of a nest-box (experimental group × ownership × 287

15

sampling day: p = 0.03). To investigate this further, the effect of treatment on plasma 288

ROMs was tested separately for owners and non-owners, using similar mixed models 289

including the treatment × sampling day interaction (Table 1, Fig. 3). In owners, there 290

was no effect of treatment or sampling day, nor of their interaction. In non-owners, 291

plasma ROMs increased in treated birds as compared to control birds (Table 1, Fig. 3). 292

There was no effect of treatment nor of ownership on body mass (Table 1); there was 293

only a significant decrease of body mass (estimate±SE: -2.32±0.46) with time. 294

The reduced model of the total song rate shows that it increased over the 295

experiment irrespective of treatment or ownership (Table 2, Fig. 4) and that it was 296

higher in owners than non-owners (Table 2). The difference in either nest-box oriented 297

or undirected song rate between control and treated individuals depended on whether a 298

bird was or was not owner of a nest-box (experimental group × ownership × sampling 299

day, p-values < 0. 001). To investigate this further, the effect of treatment on these two 300

song rate metrics was tested separately for owners and non-owners, using similar mixed 301

models including the treatment × sampling day interaction. In owners, the reduced 302

models showed that the undirected song rate increased over the experiment irrespective 303

of treatment, while the nest-box oriented song rate was stable over the experiment and 304

did not differ between treatment groups (Table 2, Fig. 4). In non-owners, the undirected 305

song rate was lower in treated birds than in control birds toward the end of the 306

experiment as shown by post-hoc tests (Fig. 4). At the same time, post-hoc tests showed 307

that the nest-box oriented song rate was significantly lower in treated birds than in 308

control birds during the first part of the experiment (Table 2, Fig. 4). 309

There was generally a negative covariation between total song rate and d-ROMs 310

values, but only the model including total song rate on day 2 was significant (Table 3). 311

16

Seven out of 12 models showed a significant positive covariation between undirected 312

song rate and d-ROMs values (Table 3). Eleven out of 12 models showed a significant 313

negative covariation between nest-box oriented song rate and d-ROMs values (Table 3). 314

315

Discussion 316

Experimental inhibition of the production of a key cellular antioxidant (glutathione) 317

enabled us to reveal a causal effect of a deregulation of oxidative balance on song 318

production. Our data also showed that the effect of the treatment on song was dependent 319

on nest-box ownership, with only non-owner birds suffering a reduction of song rate. 320

Considering that nest-box ownership reflects social dominance, our results appear to 321

indicate that subordinate males (non-owners) suffered increased oxidative damage as 322

compared to dominant males (owners). This result provides experimental support to the 323

hypothesis that the individual oxidative balance may influence the expression of 324

condition-dependent signals, hereby underlying their honesty. Analysis of individual 325

variation in song behaviour showed that a high song activity away from the nest-box 326

was associated with higher oxidative damage, while singing more at the nest-box was 327

associated with lower oxidative damage, respectively. 328

Irrespective of the ownership, our treatment prevented the upregulation of GSH 329

synthesis in all treated birds. However, only non-owner treated birds showed an 330

increased level of plasma oxidative damage metabolites. It might be that individuals of 331

high social rank were capable of upregulating their antioxidant defences to buffer any 332

oxidation induced from the inhibition of glutathione, thus avoiding any costs for the 333

production of vocalizations. On the other hand, low rank individuals were unable to 334

avoid increased oxidative damage, which came at a cost for the expression of song. 335

17

Overall, these results support the hypothesis that song rate may signal the individual 336

capacity to withstand oxidative stress (Van Hout, Eens & Pinxten 2011; Casagrande et 337

al. 2014; Baldo et al. 2015), supporting the idea that the song rate might reflect 338

individual quality (Garamszegi et al. 2004). 339

Our results indicate that dominant individuals have higher resistance to oxidative 340

stress than subordinate individuals during the non-mating season. Previous correlational 341

work on the link between social status and oxidative status in males before the start of 342

the mating season has provided controversial evidence. For example, higher-ranking 343

male rhesus macaques (Macaca mulatta) or mandrills (Mandrillus sphinx) had lower 344

levels of oxidative damage (Beaulieu et al. 2014; Georgiev et al. 2016). Conversely, 345

prior to breeding, male white-browed sparrow weavers (Plocepasser mahal) and male 346

Seychelles warblers (Acrocephalus sechellensis) did not show rank-related differences 347

in markers of oxidative damage or antioxidant protection (van de Crommenacker et al. 348

2011; Cram, Blount & Young 2015). What are the exact mechanisms via which 349

dominance status is linked to resistance to oxidative stress and why such a link varies 350

among and within species remain open questions. 351

Our experimental inhibition of glutathione caused suppression of undirected 352

song rate in non-owner birds only. The undirected song of male European starlings is 353

tightly coupled to a positive (or less negative) physiological state (Riters & Stevenson 354

2012; Kelm-Nelson & Riters 2013; Riters et al. 2014) and it is used for maintaining 355

group cohesion and social order (Eens 1997). It might be that the prolonged suppression 356

of glutathione and increased generation of oxidative damage were responsible for an 357

organism physiological deregulation that led non-owner birds to invest less in social 358

communication. Suppression of undirected song rate might also indicate higher 359

18

metabolic costs associated with it as suggested by the higher oxidative damage in those 360

males that sang more away from the nest-box. On the other hand, after an initial 361

decrease due to our experimental treatment, the nest-box-oriented song rate of non-362

owners was no longer affected by our treatment. This result suggests that resources 363

invested to regulate the oxidative balance were not taken away from song production. 364

The nest-box-oriented song rate was always negatively associated with oxidative 365

damage at individual level. The nest-box-oriented song rate is mainly used for the 366

acquisition and defense of a nest site (Gwinner, Gwinner & Dittami 1987; Eens, Pinxten 367

& Verheyen 1990; 1991). We have, for example, observed repeatedly non-owner males 368

singing on a perch of an occupied nest-box, in an attempt to challenge its owner. The 369

bird owning the nest-box reacted by singing and chasing it from the nest-box. These 370

results might indicate that non-owners prioritized investment in this song rate metric in 371

order to achieve a dominance status. Previous work conducted during the breeding 372

season also showed that, when facing an immune challenge, male starlings appear to 373

prioritize preservation of the nest-box oriented song (Costantini et al. 2015). 374

Our results also revealed temporal changes in baseline metrics of oxidative 375

balance. The concentration of GSH in red blood cells increased significantly in all 376

control individuals, while it remained stable due to the effect of sulfoximine in treated 377

birds. In both control and treated birds, concentration of GSSG increased and the 378

GSH/GSSG ratio decreased during the experiment compared to pre-manipulation 379

values, respectively. Previous studies on passerine birds also found seasonal increases 380

and decreases of GSH and GSH/GSSG ratio (Hirundo rustica, Raja-aho et al. 2012: 381

Passer domesticus, Pap et al. 2015). The reasons for seasonal changes in the glutathione 382

system (GSH, GSSG or GSH/GSSG ratio) are currently unknown. 383

19

In a previous work on male starlings (Costantini et al. 2015) it was found that 384

immunization did not affect the ratio GSH/GSSG. This is in agreement with the present 385

study, possibly suggesting that any perturbation in one of the two molecules (GSH or 386

GSSG) is somehow compensated in order to keep the ratio, and so the redox balance of 387

the cell (Jones 2005), constant. 388

389

Conclusions 390

Our study has provided evidence that experimental inhibition of an important cellular 391

antioxidant (glutathione) resulted in a decreased song rate and an increased oxidative 392

damage in males that did not own a nest-box (subordinate individuals) but not in males 393

that owned a nest-box (dominant individuals). Our results also suggest that different 394

metrics of song rate might provide different information about the individual oxidative 395

balance because they differ in how they are associated with individual quality. These 396

results provide experimental support for the hypothesis that acoustic signals may 397

honestly convey information about the individual oxidative status. 398

399

Acknowledgments 400

We thank Stefan Van Dongen and Thomas Raap for advice on statistics; Danny 401

Huybrecht for help with the HPLC analyses; Peter Scheys and Geert Eens for helping 402

with care of birds; three anonymous reviewers for providing constructive comments on 403

our manuscript. This work was supported by a postdoctoral fellowship from FWO-404

Flanders to DC. ME and RP were supported by the University of Antwerp and FWO-405

Flanders. SM was supported by a scholarship from the University of Rome "Sapienza".. 406

20

This study was done in agreement with the Belgian and Flemish legislation and was 407

approved by the ethical committee of the University of Antwerp (code 2013-28). 408

409

References 410

Baldo, S., Mennill, D.J., Guindre-Parker, S., Gilchrist, H.G. & Love O.P. (2015) The 411

oxidative cost of acoustic signals: examining steroid versus aerobic activity 412

hypotheses in a wild bird. Ethology, 121, 1081–1090. 413

Ball, G.F., Sockman, K.W., Duffy, D.L. & Gentner T.Q. (2006) A neuroethological 414

approach to song behavior and perception in European starlings: interrelationships 415

among testosterone, neuroanatomy, immediate early gene expression, and immune 416

function. Advances in the Study of Behavior, 36, 59–121. 417

Beaulieu, M., Mboumba, S., Willaume, E., Kappeler, P.M. & Charpentier M.J.E. (2014) 418

The oxidative cost of unstable social dominance. Journal of Experimental 419

Biology, 217, 2629-2632. 420

Casagrande, S., Pinxten, R., Zaid, E. & Eens M. (2014) Carotenoids, birdsong and 421

oxidative status: administration of dietary lutein is associated with an increase in 422

song rate and circulating antioxidants (albumin and cholesterol) and a decrease in 423

oxidative damage. PLoS ONE, 9, e115899. 424

Casagrande, S., Pinxten, R., Zaid, E. & Eens M. (2015) Birds receiving extra 425

carotenoids keep singing during the sickness phase induced by inflammation. 426

Behavioral Ecology and Sociobiology, 69, 1029–1037. 427

Casagrande, S., Pinxten, R. & Eens M. (2016) Honest signaling and oxidative 428

stress: the special case of avian acoustic communication. Frontiers in Ecology 429

and Evolution, 4, 52. 430

21

Cordes, M.A., Stevenson, S.A. & Riters L.V. (2014) Status-appropriate singing 431

behavior, testosterone and androgen receptor immunolabeling in male European 432

starlings (Sturnus vulgaris). Hormones and Behavior, 65, 329–339. 433

Costantini, D. & Verhulst, S. (2009) Does high antioxidant capacity indicate low 434

oxidative stress? Functional Ecology, 23, 506–509. 435

Costantini, D., Casagrande, S., Casasole, G., AbdElgawad, H., Asard, H. & Pinxten, R. 436

Eens M. (2015) Immunization reduces vocal communication but does not 437

increase oxidative stress in a songbird species. Behavioral Ecology and 438

Sociobiology, 69, 829–839. 439

Costantini, D., Casasole, G., AbdElgawad, H., Asard, H. & Eens M. (2016) 440

Experimental evidence that oxidative stress influences reproductive decisions. 441

Functional Ecology, 30, 1169-1174. 442

Cram, D.L., Blount, J.D. & Young, A.J. (2014) Oxidative status and social dominance 443

in a wild cooperative breeder. Functional Ecology, 29, 229–238. 444

Crawley, M.J. (2007) The R Book. John Wiley & Sons, West Sussex, UK. 445

De Ridder, E., Pinxten, R. & Eens, M. (2000) Experimental evidence of a testosterone-446

induced shift from paternal to mating behaviour in a facultatively polygynous 447

songbird. Behavioral Ecology and Sociobiology, 49, 24–30. 448

DeVries M.S., Cordes M.A., Rodriguez J.D., Stevenson S.A. & Riters L.V. (2016) 449

Neural endocannabinoid CB1 receptor expression, social status, and behavior in male 450

European starlings. Brain Research, 1644, 240–248. 451

Dreiss, A.N., Navarro, C., de Lope, F. & Møller A.P. (2008) Effects of an immune 452

challenge on multiple components of song display in barn swallows Hirundo rustica: 453

implications for sexual selection. Ethology, 114, 955–964. 454

22

Duffy, D.L. & Ball, G.F. (2002) Song predicts immunocompetence in male European 455

starlings (Sturnus vulgaris). Proceedings of the Royal Society of London B, 269, 456

847–852. 457

Eens, M., Pinxten, R. & Verheyen, R.F. (1990) On the function of singing and wing-458

waving in the European starling Sturnus vulgaris. Bird Study, 37, 48–52. 459

Eens, M., Pinxten, R. & Verheyen, R.F. (1991) Male song as a cue for mate choice in 460

the European starling. Behaviour, 116, 210–238. 461

Eens, M. (1997) Understanding the complex song of the European starling: an 462

integrated ethological approach. Advances in the Study of Behavior, 26, 355–434. 463

Galván, I. & Alonso-Alvarez, C. (2008) An intracellular antioxidant determines the 464

expression of a melanin-based signal in a bird. PLoS ONE, 3, e3335. 465

Garamszegi, L.Z., Moller, A.P., Torok, J., Michl, G., Peczely, P. & Richard, M. (2004) 466

Immune challenge mediates vocal communication in a passerine bird: an experiment. 467

Behavioral Ecology, 15, 148–157. 468

Garamszegi, L.Z. & Eens, M. (2004) Brain space for a learned task: strong intraspecific 469

evidence for neural correlates of singing behavior in songbirds. Brain Research 470

Reviews, 44, 187-193. 471

Garratt, M. & Brooks, R.C. (2012) Oxidative stress and condition-dependent sexual 472

signals: more than just seeing red. Proceedings of the Royal Society of London B, 473

279, 3121–3130. 474

Georgiev, A.V., Muehlenbein, M.P., Prall, S.P., Thompson, M.E. & Maestripieri D. 475

(2016) Male quality, dominance rank, and mating success in free-ranging rhesus 476

macaques. Behavioral Ecology, 10.1093/beheco/arv008. 477

23

Gil, D. & Gahr, M. (2002) The honesty of bird song: multiple constraints for multiple 478

traits. Trends in Ecology and Evolution, 17, 133–140. 479

Griffith, O.W. & Meister, A. (1979) Potent and specific inhibition of glutathione 480

synthesis by buthionine sulfoximine (S-n-butyl homocysteine sulfoximine). Journal 481

of Biological Chemistry, 254, 7558–7560. 482

Griffith, O.W. (1982) Mechanism of action, metabolism, and toxicity of buthionine 483

sulfoximine and its higher homologs, potent inhibitors of glutathione synthesis. 484

Journal of Biological Chemistry, 257, 13704-13712. 485

Gwinner, H., Gwinner, E. & Dittami, J. (1987) Effects of nestboxes on LH, 486

testosterone, testicular size and the reproductive behavior of male European starlings 487

in spring. Behaviour, 103, 68–82. 488

Gwinner, H., Van’t Hof, T. & Zeman, M. (2002) Hormonal and behavioral responses of 489

starlings during a confrontation with males or females at nest boxes during the 490

reproductive season. Hormones and Behavior, 42, 21–31. 491

Halliwell, B.H. & Gutteridge, J.M.C. (2007) Free radicals in biology and medicine. 492

Oxford University Press, Oxford. 493

Harrison, X.A. (2014) Using observation-level random effects to model overdispersion 494

in count data in ecology and evolution. PeerJ, 2, e616. 495

Hill, G.E. (2011) Condition-dependent traits as signals of the functionality of vital 496

cellular processes. Ecology Letters, 14, 625–634. 497

Humfeld, S.C. (2013) Condition-dependent signaling and adoption of mating tactics in 498

an amphibian with energetic displays. Behavioral Ecology, 24, 859–870. 499

24

Hunt, J., Brooks, R., Jennions, M.D., Smith, M.J., Bentsen, C.L. & Bussière, L.F. 500

(2004) High-quality male field crickets invest heavily in sexual display but die 501

young. Nature, 432, 1024–1027. 502

Jones, D.P. (2006) Redefining oxidative stress. Antioxidant & Redox Signaling, 8, 503

1865–1879. 504

Kelm-Nelson, C.A., Forbes-Lorman, R.M., Auger, C.J. & Riters, L.V. (2011) Mu-505

opioid receptor densities are depleted in regions implicated in agonistic and sexual 506

behavior in male European starlings (Sturnus vulgaris) defending nest sites and 507

courting females. Behavioural Brain Research, 219, 15–22. 508

Kelm-Nelson, C.A., Stevenson, S.A. & Riters, L.V. (2012) Context-dependent links 509

between song production and opioid-mediated analgesia in male European starlings 510

(Sturnus vulgaris). PLoS ONE, 7, e46721. 511

Kelm-Nelson, C.A. & Riters, L.V. (2013) Curvilinear relationship between mu-opioid 512

receptor labeling and undirected song in male European starlings (Sturnus vulgaris). 513

Brain Research, 1527, 29–39. 514

Koch, R.E. & Hill, G.E. (2016) An assessment of techniques to manipulate oxidative 515

stress in animals. Functional Ecology, doi: 10.1111/1365-2435.12664. 516

Koren, L. (2006) Vocalization as an indicator of individual quality in the rock hyrax. 517

Ph.D. Thesis, Department of Zoology, Tel-Aviv University, Israel. 518

Lindström, J., Pike T.W., Blount, J.D. &Metcalfe, N.B. (2009) Optimisation of resource 519

allocation can explain the temporal dynamics and honesty of sexual signals. The 520

American Naturalist, 174, 515–525. 521

25

MacDougall-Shackleton, S.A., Dindia, L., Newman, A.E.M., Potvin, D., Stewart, K.A. 522

& MacDougall-Shackleton, E.A. (2009) Stress, song and survival in sparrows. 523

Biology Letters, 5, 746–748. 524

Martin, P. & Bateson, P. (2007) Measuring behaviour: an introductory guide. 525

Cambridge University Press, Cambridge, UK. 526

Pap, P.L., Pătraş, L., Osváth, G., Buehler, D.M., Versteegh, M.A., Sesarman, A., 527

Banciu, M. & Vágási, C.I. (2015) Seasonal patterns and relationships among 528

coccidian infestations, measures of oxidative physiology, and immune function in 529

free-living house sparrows over an annual cycle. Physiological and Biochemical 530

Zoology, 88, 395–405. 531

Pike, T.W., Blount, J.D., Lindström, J. & Metcalfe, N.B. (2010) Dietary, carotenoid 532

availability, sexual signalling and fuctional fertility in sticklebacks. Biology Letters, 533

6, 191–193. 534

Pinxten, R., De Ridder, E., Balthazart, J. & Eens M. (2002) Context-dependent effects 535

of castration and testosterone treatment on song in male European starlings. 536

Hormones and Behavior, 42, 307–318. 537

R Core Team (2013) R: A Language and Environment for Statistical Computing. R 538

Foundation for Statistical Computing, Vienna, Austria. 539

Raja-aho, S., Kanerva, M., Eeva, T., Lehikoinen, E., Suorsa, P., Gao, K., Vosloo, D. & 540

Nikinmaa, M. (2012) Seasonal variation in the regulation of redox state and some 541

biotransformation enzyme activities in the barn swallow (Hirundo rustica L.). 542

Physiological and Biochemical Zoology, 85. 148–158. 543

Riters, L.V. & Stevenson, S.A. (2012) Reward and vocal production: song-associated 544

place preference in songbirds. Physiology & Behavior, 106, 87–94. 545

26

Riters, L.V., Eens, M., Pinxten, R., Duffy, D.L., Balthazart, J. & Ball, G.F. (2000) 546

Seasonal changes in courtship song and the medial preoptic area in male European 547

starlings (Sturnus vulgaris). Hormones and Behavior, 38, 250–261. 548

Riters, L.V., Stevenson, S.A., DeVries, M.S. & Cordes, M.A. (2014) Reward associated 549

with singing behavior correlates with opioid-related gene expression in the medial 550

preoptic nucleus in male European starlings. PloS ONE, 9, e115285. 551

Ritschard, M. & Brumm, H. (2012) Zebra finch song reflects current food availability. 552

Evolutionary Ecology, 26, 801–812. 553

Romero-Haro, A.A. & Alonso-Alvarez, C. (2015) The level of an intracellular 554

antioxidant during development determines the adult phenotype in a bird species: a 555

potential organizer role for glutathione. American Naturalist, 185, 390–405. 556

Sartor, J.J. & Ball, G.F. (2005) Social suppression of song is associated with a reduction 557

in volume of a song-control nucleus in European starlings (Sturnus vulgaris). 558

Behavioral Neuroscience, 119, 233–244. 559

Sinha, A.K., AbdElgawad, H., Giblen, T., Zinta, G., De Rop, M., Asard, H., Blust, R. & 560

De Boeck, G. (2014) Anti-oxidative defences are modulated differentially in three 561

freshwater teleosts in response to ammoniainduced oxidative stress. PLoS ONE, 9, 562

e95319. 563

Sohal, R.S. & Orr, W.C. (2012) The redox stress hypothesis of aging. Free Radical 564

Biology & Medicine, 52, 539–555. 565

Spencer, K.A., Buchanan, K.L., Goldsmith, A.R. & Catchpole C.K. (2004) 566

Developmental stress, social rank and song complexity in the European starling 567

(Sturnus vulgaris). Biology Letters, 271, 121–123. 568

27

Van de Crommenacker, J., Komdeur, J., Burke, T. & Richardson, D.S. (2011) Spatio-569

temporal variation in territory quality and oxidative status: a natural experiment in 570

the Seychelles warbler (Acrocephalus sechellensis). Journal of Animal Ecology, 80, 571

668–680. 572

Van Hout, A.J.M., Eens, M., Balthazart, J. & Pinxten, R. (2009) Complex modulation 573

of singing behavior by testosterone in an open-ended learner, the European Starling. 574

Hormones and Behavior, 56, 564–573. 575

Van Hout, A.J.M., Pinxten, R., Darras, V.M. & Eens, M. (2012) Testosterone increases 576

repertoire size in an open-ended learner: an experimental study using adult male 577

European starlings (Sturnus vulgaris). Hormones and Behavior, 62, 563-568. 578

Van Hout, A.J.M., Eens, M. & Pinxten, R. (2011) Carotenoid supplementation 579

positively affects the expression of a non-visual sexual signal. PLoS ONE, 6, e16326. 580

von Schantz, T., Bensch, S., Grahn, M., Hasselquist, D. & Wittzell, H. (1999) Good 581

genes, oxidative stress and condition- dependent sexual signals. Proceedings of the 582

Royal Society of London B, 266, 1–12. 583

York, J.E., Radford, A.N., Groothuis, T.G. & Young A.J. (2016) Dominant male song 584

performance reflects current immune state in a cooperatively breeding songbird. 585

Ecology and Evolution, 6, 1008–1015. 586

587

28

Table 1. Outcomes of both full and reduced linear mixed models of factors affecting 588

reduced glutathione, oxidised glutathione, the GSH/GSSG ratio, reactive oxygen 589

metabolites and body mass of starlings. 590

Full model

Reduced model

Variable Effect d.f. F P d.f. F P

Reduced glutathione (GSH)

Treatment group 1,27 15.65 <0.001 1,28 16.28 <0.001

Ownership 1,27 0.64 0.43 1,28 0.67 0.42

Sampling day 1,27 6.66 0.02 1,29 6.69 0.02

Treatment group × ownership

1,27 0.05 0.82

Treatment group × sampling day

1,27 7.72 0.01 1,29 7.72 0.01

Ownership × sampling day 1,27 0.01 0.92

Treatment group × ownership × sampling day

1,27 3.45 0.07

Oxidised glutathione (GSSG)

Treatment group 1,54 0.03 0.87 1,58 0.02 0.88

Ownership 1,54 0.19 0.67 1,58 0.17 0.68

Sampling day 1,54 21.4 <0.001 1,58 22.67 <0.001

Treatment group × ownership

1,54 0.40 0.53

Treatment group × sampling day

1,54 0.04 0.83

Ownership × sampling day 1,54 1.07 0.31

Treatment group × ownership × sampling day

1,54 0.01 0.92

GSH/GSSG Treatment group 1,54 1.56 0.22 1,58 1.60 0.21

Ownership 1,54 1.38 0.25 1,58 1.41 0.24

Sampling day 1,54 9.45 0.003 1,58 9.87 0.003

Treatment group × ownership

1,54 0.28 0.60

Treatment group × sampling day

1,54 0.002 0.97

Ownership × sampling day 1,54 0.001 0.98

Treatment group × ownership × sampling day

1,54 0.62 0.44

Reactive oxygen metabolites owners

Treatment group 1,13 0.07 0.80 1,13 0.07 0.80

Sampling day 1,14 1.70 0.21 1,15 1.80 0.20

Treatment group × sampling day

1,14 0.18 0.68

Reactive oxygen metabolites non-owners

Treatment group 1,13 6.92 0.02 1,13 6.92 0.02

Sampling day 1,13 1.63 0.22 1,13 1.63 0.22

Treatment group × sampling day

1,13 7.39 0.02 1,13 7.39 0.02

Body mass Treatment group 1,26 0.21 0.65 1,27 0.22 0.64

Ownership 1,26 0.51 0.48 1,27 0.54 0.47

Sampling day 1,27 25.13 <0.001 1,30 27.15 <0.001

Treatment group × ownership

1,26 0.03 0.88

Treatment group × sampling day

1,27 0.30 0.59

Ownership × sampling day 1,27 0.64 0.43

Treatment group × ownership × sampling day

1,27 0.02 0.90

591

592

29

Table 2. Outcomes of both full and reduced generalized linear mixed models with a 593

binomial error distribution showing the factors affecting total song rate, undirected song 594

rate and nest-box oriented song rate of starlings. 595

596

Full model Reduced model

Variable Effect z p z p

Total song rate Treatment group -0.61 0.54 -1.07 0.29

Ownership 0.39 0.69 2.0 0.045

Sampling day 2.49 0.013 4.17 <0.001

Treatment group × ownership 0.46 0.64

Treatment group × sampling day -1.12 0.26

Ownership × sampling day -0.31 0.76

Treatment group × ownership × sampling day 1.03 0.31

Undirected song rate owners

Treatment group -1.07 0.29 -0.22 0.83

Sampling day 2.0 0.046 4.33 <0.001

Treatment group × sampling day 1.56 0.12

Nest-box oriented song rate owners

Treatment group 1.07 0.28 0.56 0.57

Sampling day 0.26 0.80 -0.77 0.44

Treatment group × sampling day -1.12 0.26

Undirected song rate non-owners

Treatment group 1.47 0.14 1.47 0.14

Sampling day 4.14 <0.001 4.14 <0.001

Treatment group × sampling day -3.56 <0.001 -3.56 <0.001

Nest-box oriented song rate non-owners

Treatment group -3.35 <0.001 -3.35 <0.001

Sampling day -2.9 0.004 -2.9 0.004

Treatment group × sampling day 3.5 <0.001 3.5 <0.001

597

598

30

Table 3. Outcomes of linear mixed models showing the association between d-ROMs 599

values (oxidative damage) and each metric of song rate. In each model, pre- and post-600

treatment values of d-ROMs were pooled together. Pre-treatment values of each song 601

rate metric are those measured on day 1, i.e., before the start of the treatment (see Fig. 602

1). As post-treatment values of each song rate metric, we used data collected on day 2, 603

3, 4, 5, 6, 7, 8, 9, 10, 11, 12, or 13, respectively. Significant associations are shown in 604

bold time. SE = standard error 605

606

Total song rate Undirected song

rate

Nest-box

oriented song

rate

Day estimate±SE p-value estimate±SE p-value estimate±SE p-value

2 -0.35±0.16 0.027 0.25±0.12 0.048 -0.57±0.12 <0.001

3 -0.02±0.13 0.90 0.29±0.13 0.035 -0.28±0.13 0.029

4 -0.15±0.13 0.24 0.36±0.14 0.015 -0.48±0.13 <0.001

5 -0.07±0.12 0.54 0.31±0.13 0.017 -0.39±0.11 0.001

6 -0.08±0.16 0.62 0.38±0.16 0.024 -0.46±0.13 <0.001

7 0.03±0.17 0.86 0.51±0.15 0.002 -0.48±0.14 0.001

8 -0.07±0.15 0.64 0.23±0.14 0.11 -0.31±0.12 0.014

9 -0.20±0.15 0.18 0.18±0.14 0.21 -0.33±0.13 0.009

10 -0.32±0.17 0.068 0.23±0.17 0.18 -0.51±0.16 0.002

11 -0.18±0.19 0.34 0.15±0.15 0.33 -0.32±0.14 0.022

12 -0.21±0.15 0.18 0.17±0.14 0.21 -0.41±0.15 0.006

13 0.03±0.17 0.85 0.39±0.19 0.041 -0.29±0.15 0.053

607

608

31

Figure captions 609

Figure 1. Timeline of the experiment. BM = body mass. The average values of song rate 610

scores recorded from 12 to 21 November were used as pre-treatment values. 611

612

Figure 2. Pre- and post-manipulation levels of reduced (GSH) and oxidized (GSSG) 613

glutathione and the ratio GSH/GSSG in our control (C) and treated (T) male European 614

starlings (Sturnus vulgaris). The experimental treatment with sulfoximine (inhibitor of 615

glutathione synthesis) was able to prevent an upregulation of GSH. Means that do not 616

share the same letter are significantly different from each other. Data are shown as mean 617

± standard error. 618

619

Figure 3. Pre- and post-manipulation levels of plasma reactive oxygen metabolites 620

(marker of oxidative damage) of male European starlings (Sturnus vulgaris) in relation 621

to treatment (C = control, T = treated) and nest-box ownership (N = non-owner of a 622

nest-box, Y = owner of a nest-box). The experimental treatment with sulfoximine 623

(inhibitor of glutathione synthesis) caused an increase of reactive oxygen metabolites in 624

non-owner treated birds only. Means that do not share the same letter are significantly 625

different from each other. Data are shown as mean ± standard error. 626

627

Figure 4. Temporal trends of song rate metrics throughout the experiment in relation to 628

treatment (C = control, T = treated) and nest-box ownership (N = non-owner, Y = 629

owner) in male European starlings (Sturnus vulgaris). Pre-treatment (i.e., sampling day 630

1) values of each song metric are expressed as the average of the values recorded from 631

12 to 21 November. The (*) indicates a difference statistically significant. Lines that 632

32

join the box-plots are shown when there are differences statistically significant. Data are 633

shown as mean ± standard error. 634

635

33

636

Figure 1 637

638

Blood and BM

First Injection

Song Song Song

Second Injection

Song Song

Third Injection

12 to 21 Nov.

Song

Fifth Injection

23 Nov. 24 Nov. 25 Nov. 26 Nov. 27 Nov.

Song

Fourth Injection

Song Song

28 Nov. 29 Nov. 30 Nov. 01 Dec.

Song Song

Blood and BM

02 Dec. 03 to 05 Dec.

34

639

Figure 2 640

641

Oxid

ise

d g

luta

thio

ne

GS

SG

(µm

ol g

-1 f

resh

weig

ht

of

red

blo

od c

ells

)

0

0.3

0.6

0.9

1.2

0

0.5

1.0

1.5

2.0

Red

uced

glu

tath

ion

e G

SH

(µm

ol g

-1 f

resh w

eig

ht

of

red

blo

od c

ells

)

0

1.0

2.0

3.0

4.0

GS

H/G

SS

G

a a b a

Pre-treatment values Post-treatment values

controls treated

35

642

Figure 3 643

644

Re

active

oxyg

en

me

tabo

lite

s

(mM

H2O

2 e

quiv

ale

nts

) 0.6

0.5

0.4

0.3

0.2

0.6

0.5

0.4

0.3

0.2

a a a b

Pre-treatment values Post-treatment values

No

n-o

wn

ers

O

wn

ers

controls treated

36

645

Figure 4 646

Sampling day

No

n-o

wn

ers

To

tal so

ng

ra

te (

%)

0

10

20

30

40

50

60

0

10

20

30

40

50

60

0

10

20

30

40

0

10

20

30

40

Ne

st-

bo

x o

rie

nte

d s

on

g r

ate

(%

)

* *

Un

dire

cte

d s

on

g r

ate

(%

)

0

10

20

30

40

50

60

0

10

20

30

40

50

60* * *

Ow

ne

rsN

on-o

wn

ers

Ow

ne

rsN

on-o

wn

ers

Ow

ne

rs

1 2 3 4 5 6 7 8 9 10 11 12 13

controls treated

* * *